This message was deleted.

Hey! So yeah, nothing crazy expected and it should just work. We have specific plugins for geopandas, but they only exist as geopandas subclassed pandas. See https://github.com/DAGWorks-Inc/hamilton/blob/98471b7b1592486c3eed11c0300f65510bf2de55/hamilton/plugins/geopandas_extensions.py#L20. If you do hit anything, building a custom plugin like geopandas should be easy and will likely cover many of the cases — the touch-points/coupling between pandas (+ polars, dask, spark, etc…) and Hamilton are done largely through similar plugins. If for some reason that doesn’t cover it, we’d be happy to dig in and help find/fix in Hamilton 🙂

As far as plugins go, that’s pretty short. Thank you very much.

Yep! Its a simple one 🙂 I don’t think you’ll need to use them but happy to walk you through plugins/how we’ve organized them if you do want to add your own (either for local running or sharing back)

One more thing - you might want to come up with custom code like https://github.com/DAGWorks-Inc/hamilton/blob/98471b7b1592486c3eed11c0300f65510bf2de55/hamilton/plugins/h_polars.py to create a final object/result.

For reference what we’re considering would be very similar to the Pint Pandas extension https://github.com/hgrecco/pint-pandas/tree/master

Yeah, given that that operates at a lower level and you’re not subclassing I don’t think you’ll have to deal with plugins at all

Awesome news

Yeah should be easy to validate.

Thank you both